Impaired judgment and decision making is a key factor contributing to drug abuse. Research on the source of these impairments has focused largely on cognitive and neural deficits during two distinct periods. Before making a choice, drug abusers are overly sensitive to potential rewards and insensitive to long-term losses. After making a choice, drug abusers are hyper-reactive to rewards and show poor ability to learn from experience. Much less is known about the intervening process of deliberation when beliefs and desires are integrated over time to form a preference leading to a choice. This deliberation process can determine whether a person appears risk seeking or risk averse, impulsive or cautious, and slow or fast in responding. While these are characteristics directly relevant to drug abuse, the basic mechanisms of the process are not well understood. We conceptualize deliberation as a sequential sampling process where decision makers evaluate possible payoffs forming a subjective valence. These valences are accumulated over time forming a preference over the alternatives until a threshold is reached initiating a choice. In this application, we develop a theoretical and experimental framework that integrates computational modeling and cognitive neuroscience to characterize this deliberation process. Experimentally, we create a novel gambling task called the flash gambling task (FGT) in which participants choose between a sure payoff and a lottery that offers a draw from a distribution of payoffs. Instead of receiving verbal descriptions of the lottery, subjects watch simulated draws from this lottery that flash by every 50 ms (like watching a stock ticker run by). Thus, the FGT requires active integration of payoffs allowing more precise control over the deliberation process. Theoretically, we develop a framework that integrates computational models of decision making, neural studies of reward processing and perceptual decision making, and analytic models of hemodynamic response. Our model makes specific predictions regarding the fMRI BOLD signatures of different aspects of deliberation process during risky decision making. In this application, we use this model-based imaging approach to delineate the neural circuitry underlying the valuation and preference formation process in an fMRI experiment on a normal college sample. In a second study, we investigate the link between behavior and deliberation in the FGT and measures of risky drug use and impulsivity using a larger community sample. Results from these studies will offer new insights on the basic cognitive and neural mechanisms of risky decision making and establish potentially important links between process-level measures of choice behavior and drug use, thereby setting the stage for a greater understanding of the neural and computational basis of drug abuse.